Quantitatively measure the impact of DNA vs epigenetics?

Question

Background: While classical genetics deals with how DNA sequences directly code for traits, epigenetics involves chemical modifications to DNA and associated proteins that can switch genes on or off without changing the underlying DNA sequence (including DNA methylation, histone modifications, and chromatin accessibility, etc).

Question: Is it possible to quantitatively measure and compare the impact of DNA and epigenetics on expression? Which one is more important and by how much? If so,

What is the proportional contribution of epigenetic modifications versus DNA sequence variation to gene expression variance in human somatic cells, expressed as percentages of total phenotypic variance?

Example equation:

$\text{Total Phenotypic Variance} = \sigma^2_G + \sigma^2_E + \sigma^2_{G \times E} + \sigma^2_{\text{error}}$

Here, G means genetics, E means epigenetics, and the third term is covariance.

Example result: I wonder if there are already results like this: Genetic sequence variation explains about 25-40% of expression variance in a sample. Epigenetic variation accounts for approximately 10-30% of expression variance. The remaining variance is attributed to environmental factors, measurement noise, and gene-environment interactions

There are surely a lot of loopholes and drawbacks in this approach, so I wonder if there are better approaches. I think this quantitative analysis would be very important, because it can guide us in many areas. For example, in genetic engineering, we need to know if it is more efficient to edit the gene or to deal with the epigenetics.

Answer 1

There is extensive work attempting to decompose phenotypic variance into genetic, epigenetic etc. terms. Here are just three random examples from the last 10 years that I got by googling this:

• https://www.nature.com/articles/s41437-019-0261-8
• https://www.nature.com/articles/s41467-020-16520-1 (focuses specifically on DNA methylation, which is possibly dubious from a classic interpretation of epigenetics)
• https://www.nature.com/articles/s41437-018-0114-x.pdf (theoretical review)

In some cases phenotypes show >50% of variance could be explained by epigenetic causes (see review). In other cases it is much less.

However, as always when considering these variance decomposition approaches, variance decomposition does not substitute for a causal analysis. Heritability is far too unstable of an estimator to be informative except in very well designed cases- that are more or less impossible to find for humans, because we can't breed humans for ethical reasons.

Epigenetic analysis only complicates this ambiguous approach, given its intimate relationship to both genetic and environmental contributors to phenotype.